GBC* is an established conference organizer for the downstream. The IIOT & Digital Solutions for Oil & Gas event held in Amsterdam last month was its first venture upstream. Kari Jordan provided a keynote introduction to Shell’s Tacit IT initiative. Digitalization is changing the energy industry and helping the world decouple global emissions from economic growth, notably with the internet of things, machine learning and big data/analytics (BDA). Shell has many IoT experiments running and a small team working on cognitive computing. Shell’s data represents a ‘huge untapped source of richness.’ Shell is running robotics proof-of-concept trials and also has a team working on blockchain. Jordan thinks that blockchain, which has no central authority, ‘may change current thought process around business models.’ Shell’s iScope virtual reality geoscience center got a plug as did Honeywell’s operator training simulator for the Prelude platform. Analytics in Shell is ‘quite mature,’ but there are still questions as to how to incorporate these developments in future plants.
McKinsey’s Sverre Fjeldstad made the bold claim that digital can be used to ‘fix the economics of oil and gas,’ whose business model ‘has been destroyed in the last decade.’ Since the 2015 downturn, companies have done an excellent job of restoring margins by squeezing suppliers, making employees run faster and by ‘losing the gold plating.’ But this is not enough in a lower for longer oil price scenario. In particular, shale oil ‘has not been profitable since 2014.’ Future progress will come from digital and advanced analytics in predictive maintenance and ‘a lot of other use cases.’ McKinsey has applied BDA to several years of control room parameter historical data. Machine learning was applied to a 500GB data set to identify five ‘themes’ of settings that made for periods of high production. Predictive models promise a production hike of ‘1-2%.’ Having said this, Fjeldstad observed that plugging real time/ad hoc analytics back into operations was ‘a few years away!’
The conference theme, the Industrial Internet evokes standardization. Stephen Mellor of the Industrial Internet Consortium (IIC) opined that while it was relatively easy to standardize the bottom of stack, the hard part is to standardize higher up in the stack, above operating systems and communications protocols. The search is on for commonalities across verticals where there is a need for global standards. Unfortunately, ‘companies like to embrace, then extinguish their own standards as do nations.’ There are different international IIoT standards across the US (NIST), Japan (UiT acceleration consortium), China (CAICT) and the EU (AIOTI). Mellor offered the IIC’s Track & Trace testbed as an example of what could be achieved. This involved a joint venture between Airbus and Bosch to leverage IoT-enabled tool tags such that less time was wasted looking for tools! Having said that, Mellor revealed that ‘the IIC is not a standards organization!’ The IIC will ‘establish frameworks, evaluate existing standards, identify requirements and propose them to standards bodies such as ISO.’
IHS Markit’s Oscar Abbink pushed back the digital time line to the 2004 CERA White Paper that introduced the ‘digital oilfield.’ DOF program starts peaked in 2006 with the expectation of up to 20% savings in OPEX. Poster children included Kvaerner’s unmanned ‘Subsea on a Stick’ facility and real time reservoir management as practiced by Chevron in Kern River. Big data is not exactly new. There are already ‘pockets of excellence’ in the industry which may spread with the next wave of digitalization. This will see ‘convergence around a core of data-driven analytics and optimization.’
Einar Landre traced Statoil’s GoDigital program that kicked off early in 2016. GoDigital’s premise was that there were ‘threats and opportunities’ in the digital space which could impact unsustainable, post-downturn margins. This led to the creation of a digital center of excellence in March 2017. The CoE is now operational and ready to deliver the digital roadmap. The roadmap derived from 38 ‘lighthouses,’ idealized future states that provide direction. GoDigital uses a bottom-up approach, ‘top-down involves more politics than you need.’ The next step is to select a technical platform for BDA and to ‘develop a new mindset for an unmanned world.’ Statoil wants to stay profitable, even at $30 oil.
Shahrul Rashid and Jefferi Bin Kamarudin presented Petronas’ digital roadmap for the downstream. Today there may be up to 45,000 instruments in a refinery, keeping operations safe. Petronas’ refinery of the future program envisages an integrated operations and maintenance center with data captured to the Petronet cloud database. A Nagios-based dynamic risk analysis and early warning system was co-developed with Near Miss Management also ran. ‘dSCE,’ a digital toolkit for operations and maintenance leverages Petronas’ pervasive wireless network and handheld devices providing information at workers’ fingertips.
Asger Klindt presented Maersk Drilling’s work towards a common IIoT platform to support a digital twin of a drilling rig to support predictive maintenance and drilling productivity enhancement. Maesrk could not find an off-the-shelf IoT solution and so has engaged with GE to develop a solution around Predix and SmartSignal. Predix is not suited to offshore low bandwidth Vsat communications so Maersk is working on ‘edge’ analytics on the rig. Earlier this year Maersk was reported to have deployed Maana’s knowledge graph at its shipping business.
Hatem Oueslati showed how Baker Hughes (now a GE company) is using real time downhole data in its Remote Operations Services. Here, cross functional teams compare real time and historical data and analytics. A modern rig can produce over a terabyte/day. Big data support leverages Cassandra along with Energistic’s Witsml, ETP and global log curve mnemonics. Data analysis and performance benchmarking is performed with the BHI Signals platform.
While others presented somewhat high-level approaches to the IIoT, Greg Harris’s presentation of CD Power’s use of Movus FitMachine was more pragmatic. CD Power supplies electricity generators to onshore drillers. These are now monitored with Movus’ FitMachine, a hockey puck sized, self-powered magnetic device that transmits vibration and temperature into Movus’ Google/TensorFlow-powered AI engine. Machine behavior is learned in a couple of weeks from which time, the system issues alerts if machine health deteriorates. Movus’s ‘zero IT’ footprint receives software updates over the wifi network. Five have been fitted onto CDP LNG compressors too.
What follows is a narrative derived from the panel discussion that wound up day one of the conference. A potential issue raised regarding BDA was that ‘data scientists become insiders and get to know more about the business than anyone!’ In a similar vein, BDA is more of a human problem than a technological one, if you want to access say HR data, the first question you will be asked is ‘why?’ Nobody wants to relinquish control over the data asset and such reticence to share information is ‘a major cultural roadblock to change.’ On the HR front, the risk of a shortage of data scientists recalled the scramble for cyber security specialists of a couple of years back. There again, there are a lot of smart math-trained people in oil and gas that could fill data scientist roles. Data ownership is also a vexed issue. ‘If you have a GE turbine, is the data yours or GE’s? For a major operator, the data is a part of its asset, even though it may be shared with GE such that they can improve and service the machine. Cyber security is getting increasingly risky as ‘all the data eggs are in the same data lake/basket’ instead of being spread across multiple legacy systems.
Christophe Romatier presented Honeywell’s Connected Plant, delivered from its UOP unit, a century old line of business that now ‘makes over 60% of the world’s gasoline.’ Like GE, Honeywell recently decided to transform itself into an industrial software company. The connected plant offering is a digital twin of the plant with a big data engine inside. Connected Plant components include a Hadoop data lake and metadata model, Spark stream processing (time series database) and Honeywell’s cloud-based data historian.
Mikhail Korolkov (Gazprom) and Aleskey Kostrov (Rostelecom) presented their ‘Open partnership for the industrial internet of things in oil and gas.’ The combination of Russia’s major gas producer (13% of world production) and the national telco are to deliver a platform for data analytics, IoT connectivity and a marketplace, starting in 2018. A Russian national cloud platform will provide connectivity to oil wells and production facilities for real time drilling monitoring, security and more. Gazprom plans to use the platform in future operations, starting with facility design in a virtual space. The model will then run as a prototype to check performance. The system is to support plug and play construction, model-based production optimization and self-diagnosing equipment. Target architecture centers on the digital twin. Korolkov observed that although Honeywell has a lot of answers, there is also a lot of hype around the digital twin. Specific use cases are still to be determined. The scope of the digital twin, how to keep it in sync with reality and its cyber security all need further thought.
In the ensuing debate, the issue of the intellectual property (IP) in the digital twin was raised. This is a recurring question and a subject of tension between owner operators and software vendors, curiously since, ‘Operators have no mechanism for monetizing IP.’ One major reported ‘more negative than positive examples’ of collaboration as the supplier got all the goodies and ‘left us with nothing.’ The perennial issue of information handover from construction to operations was raised. Construction may take place in a virtual space but the idea of this being carried through across handover is ‘light years from today.’ ‘What exactly is the digital twin and how can we get there realistically?’ This issue has plagued oil and gas but in one flagship rail project in the UK, ‘everyone involved has access to all the data all the time.’ The commercial world could learn a lot from this government-backed approach. Another speaker acknowledged that ‘nobody has cracked this’ stakeholders ‘need to want to share.’
The chair invited speakers to give their evaluation of the state of the art of the IIoT, with a subjective scoring of technology and take-up. A major vendor opined that all are ‘just starting.’ Most have reasonably established DOF solution with sensors and are taking their first steps in analytics. But nobody is running fully autonomous plant. Most are at stage 2 moving to 3 or 4 (out of 5). Most are just at the asset configuration stage, ‘what do I have,’ although companies in the power and other regulated sectors are ahead. The fact that majors work in silos represents both a problem and the potential for a big win. But with the downturn, budgets for research have been cut. ‘Maybe more are at 2 to 3.’ A major got a ripple of laughter when he opined that ‘vendors are also at level 2.’ ‘Everyone knows how to get there but…’
Russel Herbert reported that OSIsoft’s PI System is used on some 1.5 billion real time data streams representing 45 million boe/day. Shell alone has 7.5 million connected devices performing 100,000 calculations/minute for 15,000 users. PI is therefore central to the first wave of the digital transformation and OSIsoft intends to stay ahead in the next wave of the cloud, BDA, integrated operations and mobility. Everyone is asking ‘can’t we do more with the data?’ The answer is, yes we can! 80% of the value of analytics is at the operations level and involves data from counters, trends, pressure rising too quickly, how many times on and off. Here ‘no advanced analytics is required.’ Elsewhere (the 20%) run at the enterprise level, with the application of machine learning to time series data, blending data from maintenance, finance and maybe running in the cloud. Tagged historian data may not mean much outside of plant. This needs to be fixed upfront for digital initiatives, adding context and metadata to raw tags. All oils use multiple analytics solutions providers, SAP/Hana, Watson, Azure, and the OSIsoft Cloud. Such multiple connection points may pose a security risk. Another approach has data lakes sitting beside PI which adds complexity. Streaming analytics is hard because contextualization takes time although there are some successes. OSIsoft’s recommended approach is, unsurprisingly, to have PI at the heart of the solution with all other digital oilfield tools working off of PI. Shell, ENI and Element Analytics were cited as users/partner.
* Visit the Global Business Club and the IIoT/Digital Solutions event home page.
© Oil IT Journal - all rights reserved.